Discussions of error outliers

[godotalgorithm]

By their very nature, semiempirical models have limits to their accuracy and transferability, and their further development requires problems to be identified and fixed. This can be facilitated by the reporting of molecules that are error outliers in existing models, so that they (or an appropriate proxy) can be added to future reference data. There are pages on the old MOPAC website listing known faults in the PM6 and PM7 models, but the old website is no longer being updated and the future location of such reporting should be more interactive. I've created a "Models and data" discussion category so that this type of reporting can occur here, which should be sufficient as long as the number of such reports remains relatively low.